Gendered theories of autism have contributed to a popular impression that autism is a predominantly male phenomenon. The Extreme Male Brain (EMB) theory of autism (Baron-Cohen, 2002) posits two diametrically opposed cognitive styles, “empathizing” and “systemizing,” that correspond roughly to female and male brains, respectively. Baron-Cohen's proposal of the EMB theory relied heavily on stereotypes of masculinity and femininity, such as “girls show more concern for fairness, whereas boys share less” (Baron-Cohen, 2002, p. 249) or “Males are quicker to establish hierarchies of dominance” (p. 250).

Many of Baron-Cohen's observations were rooted in the systemic misogyny that has limited the involvement of women in the public sphere. For example, Baron-Cohen notes that math, physics, and engineering “are largely male-dominated disciplines” (p. 250), neglecting that sociological reasons beyond a tendency toward empathy might contribute to the absence of females from those fields. There is copious evidence that women both expect (Chen & Moons, 2015) and encounter increased bias and harassment in male-dominated environments and disciplines (Banchefsky & Park, 2018; Dresden et al., 2017; Lawson, 2020; Street et al., 2007). Even with gender equity efforts in science education and employment, these studies indicate that the behavior of men—not the professional preferences of women—creates hostility that limits the participation of women and gender minorities.

Despite the inherent flaws in Baron-Cohen's EMB theory, two decades of empirical research has addressed the hypothesis that autism represents the most masculine possible brain structure and function. Methods and measures have been as diverse as self-report surveys, hormone levels, digit length ratios, and functional neuroimaging, all with a focus on establishing the maleness of autistic brains of any sex. The studies based on this hypothesis have produced conflicting results.

The largest EMB study to date, reported by Baron-Cohen and colleagues (Greenberg et al., 2018), surveyed more than 670,000 individuals including 36,648 autistic respondents. Although the report acknowledged that 2.5% of the autistic respondents identified as nonbinary and that this proportion is 5.5 times that in the general population (p. 12154), the authors proceeded to analyze the results of binary-gendered individuals only. Results for the nonbinary participants were not reported. The report also consistently referred to “nonbinary sex,” even though nonbinary refers to gender, not sex. The study used the Autism Quotient (AQ; Baron-Cohen et al., 2001), Empathizing Quotient (EQ; Baron-Cohen & Wheelwright, 2004), and Systemizing Quotient (SQ; Baron-Cohen et al., 2003). All of the reported results were as predicted based on a priori assumptions about male and female differences. As predicted, non-autistic females scored higher than non-autistic males on the EQ and non-autistic males scored higher than non-autistic females on the SQ. Autistic people, both male and female, scored higher on the AQ and SQ and lower on the EQ than controls (Greenberg et al., 2018). The authors interpreted the pattern of findings to mean that autistic people show a “masculinized” profile, supporting the assumption that autism represents an extreme male brain. The exclusive use of assessment instruments developed by the senior author and colleagues limits the external validity of these findings. Furthermore, results based on survey responses cannot be extrapolated to interpretations of brain structure or function.

Some studies have attempted to integrate neuroimaging methodologies into the investigation of EMB. Kozhemiako et al. (2018) used functional magnetic resonance imaging (fMRI) to compare interhemispheric connectivity in 104 autistic females, 115 autistic males, 107 non-autistic females, and 114 non-autistic males (age range six to 26 years). They reported that females and males with ASD tended to follow the same developmental trajectory for connectivity as the typically-developing males, while typically-developing females showed a unique developmental trajectory. Kozhemiako et al. (2018) interpreted their results as supporting the EMB hypothesis. The authors argued that their results indicated the effectiveness of using developmental trajectories to study sex differences in autism, but their cross-sectional study could not have provided true developmental evidence in support of the EMB hypothesis. A longitudinal neuroimaging design would be required to obtain such evidence.

Research methods that diverge even more widely from measures of the neurological and cognitive differences inherent to autism have also been used to investigate the EMB hypothesis. For example, the ratio between the lengths of the second and fourth manual digits (2D:4D; i.e., index and ring fingers) is an established indirect measure of prenatal testosterone exposure. Males tend to have longer fourth digits than second digits, whereas both digits tend to be of equal length in females (Manning et al., 1998). However, some research suggests that the relationship between 2D:4D and hormones may not be straightforward (e.g., Putz et al., 2004). Despite this potential limitation in interpretation of the results, numerous studies have employed this measure when examining autistic individuals. A critical review of 25 peer-reviewed studies found that autistic individuals tend to have more masculine 2D:4D ratios than non-autistic individuals (Teatero & Netley, 2013). Furthermore, the meta-analyses found non-significant relationships between 2D:4D and the AQ, SQ, and EQ in directions that were consistent with the EMB hypothesis, which was interpreted as providing suggestive support for the EMB hypothesis. The authors acknowledged considerable heterogeneity in the results of the reviewed studies and numerous other limitations to the measures and the meta-analyses.

Studies contradicting the EMB hypothesis have also used a wide variety of methods, including hormone levels, 2D:4D, other biometrics, and behavioral tasks. Sharpley et al. (2017) compared age-related changes in testosterone between 136 autistic and 48 non-autistic young males (age range six to 17 years) and found no group differences. Since the EMB hypothesis suggests that autistic males would have higher testosterone levels than non-autistic males, the authors interpreted this finding as failing to support EMB.

In another study related to the EMB, Falter et al. (2008) measured 2D:4D and performance on visuospatial tasks in 28 autistic children (27 male) and 31 typically-developing children (30 male). Males typically show an advantage for the visuospatial tasks (i.e., Mental Rotation, Targeting, and Disembedding). Falter et al. (2008) found no group differences in 2D:4D ratios for the autistic children and the typically-developing children and no significant relationships between 2D:4D and visuospatial performance for either of the groups. The findings of Falter and colleagues were not consistent with the EMB predictions that male autistic children would have the best visuospatial performance and that performance would be related to 2D:4D as an indirect marker of prenatal testosterone.

Bejerot et al. (2012) examined numerous physical measures including testosterone levels, circumferences of various body parts, and 2D:4D. They reported that autistic participants had androgynous physical features regardless of participant sex. That is, the autistic female participants had more masculine characteristics than the typically-developing females, and the autistic male participants had more feminine characteristics than the typically-developing males. Bejerot et al. (2012) suggested that autism should not be characterized as a masculine disorder but as a “gender defiant” one (p. 122).

A recent review by Ridley (2019) explored theoretical challenges to the EMB theory. A major challenge raised to the EMB theory in this review was that the AQ, EQ, and SQ are all internally-validated metrics designed to describe cognitive styles in the ways that Baron-Cohen and his colleagues originally conceived them. That is, the self-report metrics used to collect information about autistic traits, empathizing, and systemizing all reflect what Baron-Cohen and colleagues think autism, empathy, systemizing, masculinity, and femininity should look like. Ridley (2019) contrasts these with externally-defined metrics, which are based on direct observation of some characteristic in the human population. Ridley suggested that the measures associated with the EMB theory could be strengthened if they reflected observations of actual human behavior, rather than participant reports about characteristics of interest to the authors who were the original source of the EMB.

A corollary of the EMB theory is the Female Protective Effect (FPE; Robinson et al., 2013). The FPE explains that females “require a greater etiologic load to manifest autistic behavioral impairment” (Robinson et al., 2013, p. 5258). That is, females who are recognizably autistic are thought to have greater genetic contributions to their autism than similarly-situated males. In establishing the case for an FPE, Robinson et al. (2013) cite an earlier study that found that when autistic trait severity was held constant, boys were still diagnosed with autism more frequently than girls (Russell et al., 2011). That is, two individuals with identical patterns of autistic traits who differed only on the basis of sex received different diagnoses. Failure to diagnosis females with autism who exhibit the same traits as males diagnosed with autism indicates that the FPE, if it exists, cannot completely explain why females are less likely to receive an autism diagnosis than males. Rather than an inherent biological difference, differences in rates of autism for males and females may be complicated by biases within the diagnostic process.

Both the EMB and FPE theories, even if true, may reflect more about the nature of autism research and diagnostic practices than they do about the nature of autism. Autism may affect more males than females, but this cannot be definitely established if the diagnostic practices are compromised because of confirmation bias toward a greater inclusion of males. Furthermore, as long as the criteria used to determine who qualifies for research participation bias the inclusion of males, sex differences in autism phenotype or incidence cannot be adequately studied.

Original source: Coburn, K.L. (2021). Spoken narratives by autistic adults of under-represented genders.